Summary: | 碩士 === 國立成功大學 === 水利及海洋工程學系 === 84 === Based on the theory developed by Hashimoto(1994) about the
application of Maximum Entropy Method on Directional Wave
Spectra Estimation, a new optimization strategy ARS(Adaptive
Random Search) is chosen to simplify the complicated procedures
and to avoid the divergence in the iteration computation. The
validity and reliability of the method are examined through
numerical simulation. The influence of the errors originated
from the data acquisition and the computation procedures is
discussed. Estimating the Directional Wave Spectra with
insufficient field wave data could be regarded as an ill-posed
problem. Maximum Entropy Principle,which was proved by
Statistics and Information theory to be an effective solution to
ill-posed problems, could be used to derive the probable
expression of directional spreading function. Furthermore, the
global optimization strategy ARS method is applied to estimate
the parameters of the spreading function. Two multi-variable
functions are chosen to examine the efficiency and the
effectiveness of the ARS strategy. An AIC criterion is applied
to determine the moderate order of the directional spreading
function. This modified approach has been proved to have the
ability to describe asymmetric and bimodal Directional spreading
by the numerical simulation. It is suggested that the operation
parameters of ARS f1 and f2, could be set to 4 and 10 in order
to save the computation time with acceptable tolerance.
|